Subject: Re: SPARCstation-5/170
To: David Laight <email@example.com>
From: Kurt J. Lidl <firstname.lastname@example.org>
Date: 07/05/2004 13:23:05
On Sun, Jul 04, 2004 at 04:19:49PM +0100, David Laight wrote:
> On Sun, Jul 04, 2004 at 11:01:37AM +0000, Matthias Scheler wrote:
> > In article <20040623123308.K26659@zill.net>,
> > Brian Hechinger <email@example.com> writes:
> > > of only there were supported FDDI cards though, ...
> > The SBus FDDI cards which Sun sold (made by a 3rd party vendor whose name
> > I forgot) were slow as hell. I once tried one of them in my girlfriend's
> > SS20 (running Solaris 9) with a fast i386 system on the other site. I only
> > got about 1.1MByte/Sec (measured with "ttcp"). The machine can deliver
> > 6MByte/Sec over the hme SBus card.
> > So I would waste too much thoughts on FDDI cards for SBus systems.
> We did a driver port for an SBus FDDI card - starting from the card
> manufacturer's  solaris driver. The original version sucked. I then
> took out all the 'buffer loaning' stuff and added a data copy to both
> the transmit and receive paths. New modified driver was much, much
The FDDI cards that Sun sold were made by a company called
Network Peripherals Inc" (NPI). The provided the drivers for
SunOS 4.x and Solaris 2.x.
The cards are capable of much better than the above quotes numbers.
BSDi did a port of the SunOS driver to the ill-fated Sparc port
of BSD/OS -- on a 150Mhz HyperSPARC machine, I could forward traffic
in one FDDI interface and out another at 44mbit/second, for sustained
periods of time.
I don't remember what the raw speed of the cards was... Probably around